Abstract

This project is a study of the U.S. housing market in 2017, mainly focusing on the index: ‘list price (or asking price) for homes listed on Zillow per square foot’ and ‘Rental price per square foot of homes listed for rent’. Aims to present the housing market from multiple dimensions and visualize the results in a reader-friendly way.

On the page ‘Exploratory’, the project shows an overview of the data of these two indicators. At the same time, it also compared the difference in selling price and rent between different types of houses, as well as the difference between the actual price and the estimated house value.

On the page trends’, the project explores the data on time level, presenting the changes in list price and rental price from the beginning of 2016 to the end of 2017 and forecasting them. Through the text analysis of the monthly report (published by Zillow), the project found the most popular topic in the housing market in 2017 and conducted further research.

As for the page geographical Trends’, this is a geographical exploration of list price and rental price.

Dataset

All the data are downloaded from or scraped from the:

1. Zillow Website
   • Zillow’s Economic Research Team collects, cleans and publishes housing and economic data from a variety of public and proprietary sources.
   • The dataset contains 82 columns. The variables include: Housing price, housing type, regional data, timestampe, etc.
2. Census Bureau Website
   • The Census Bureau is part of the U.S. Department of Commerce and the mission is to serve as the nation’s leading provider of quality data about its people and economy.
   • The dataset is about migration flows between states.

Technology

This project mainly used R language, and displayed the results in the form of Dashboard. Most charts are interactive plots, readers can click the legend and graph, selecting the elements that they want to see more detailed.

The package applied:

• Data cleaning package
  • tidyr
  • dplyr
  • data.table
  • tm
  • tibble
• Visualization package:
  • ggplot2
  • hrbrthemes
  • circlize
  • ggpubr
  • wordcloud2
  • viridis
• Flexdashboard package:
  • flexdashboard
  • plotly

The model applied:

• Sarima Time Series Prediction Model

Number of Sold house

Avg House Price Per Sqft

Avg Rental Price Per Sqft

Pct of Increasing Homes Price

Pct of Sell For Gain

Distribution of All Types House Price and Rental Price (Pre Sqft)

House Price and Rental Price By Various Types (Pre Sqft)

Overview

In 2017, more than four million houses were sold in the United States, and 94% of sellers profited from these transactions. The average annual listing house price in 2017 was $157.6 per square foot, and the average annual rent was $1.1 per square foot. For both sold and rental houses, distribution shapes of actual and estimated prices are very similar concentrate on relatively low price. And real price distribution shows more left-skewed, indicating that the actual prices are higher than their estimated prices And people who rent a house are less likely to be interested in the high-value house. (see the figure on the top left for more details). Moreover, different house types have different selling prices and rents. Compared with other house types, 1bedroom and 3bedroom houses are more expensive.(see the figure on the bottom left for more details)

House Inventory Status

Listing and Rental Price Pre Sqft by Time

Word Clouds of Monthly Report

Prediction of House Listing Price Pre Sqft, prediction is created by AutoSarima model

Listing Price Per Sqft in Geographical Level

Rate Per Sqft in Geographical Level

Cities with the highest outflows(DC, CA, HI, MA)

Cities with the highest inflows(MS, MV, AR, IN)

Insights Summary

The housing market met an inventory issue in 2017. It not only reflected on the reports: the word “inventory” became a popular term been discussed, but also reflected on inventory curve: the number of the house on inventory had a significant downward trend during 2017. Although house prices and rents had steadily increased month by month as usual. The falling inventories might accelerate this growth curve.

Furthermore, the level of house prices and rents were similar in geographical distribution. The higher the house price, the more expensive the rent. House prices and rents were also highly correlated with the flow of people. For example, in CA, it has top 1 rental price and highest outflow rate. And the top 4 states that oeople are more likely to move to were the state that had relatively low house price.